The Effect of Font Size on Reading Comprehension
Research on reading development has focused on the linguistic, cognitive, and recently, metacognitive skills children must master in order to learn to read. Less focus has been devoted to how the text itself, namely the perceptual features of the words, affects children’s learning and comprehension. In this study, we manipulated perceptual properties of text by presenting reading passages in different font sizes, line lengths, and line spacing to 100 children in the second and fifth grades. For second graders (Experiment 1), decreasing font size, as well as increasing line length, yielded significantly lower comprehension scores. Line spacing had no effect on performance. For fifth graders (Experiment 2), decreasing font size yielded higher comprehension scores, yet there were no effects for line length and line spacing. Results are discussed within a "desirable difficulty" approach to reading development.
Consider the subjective experience of a second grader reading a text, poorly photocopied and written in a small font. In contrast, imagine her reading large print, centered on the page, and subjectively easy to read. Intuitively, one may think that these perceptual and typographical factors are only related to the child’s motivation to read and will not affect her comprehension. Very little research has focused on potential developmental effects of manipulating physical properties of print (e.g., print size, font type, etc.) [1-3]. In adults, for example, it has been found that altering text presentation to a less familiar format, hence making it less perceptually fluent (words in italics) led to better memory of studied material in adults and high school students . Using a different manipulation of text, another study found that extra large letter spacing enhanced the performance of word reading in children with dyslexia . Currently, due to scarce research, all that can be concluded is that the effects of altering text presentation may differ by the specific manipulation and by population. It may also affect different aspects of reading: rate, accuracy and comprehension. In this study, we focus on an understudied question which is, can a mere manipulation of perceptual features of text enhance reading comprehension among second and fifth grade children.
Reading Development and Reading Comprehension
Developmental models of reading assume that reading is made up of component skills [4-6]. These components begin with letter-sound recognition and then proceed to decoding skills. While each component is sufficient for a time, new skills must be achieved if reading proficiency is to increase. Later components include the development of efficiency, comprehension, and the ability to integrate and synthesize materials.
Children are assumed to progress from learning about print itself to learning about the alphabet, sounds of the letters, and letter groups . During the initial period of learning to decode, in the first and second grades, the reader is "glued to the print." Reading is slow and laborious, as new readers still receive many cues about how to decode words from the letters themselves. By reading material that has familiar content and language style, children develop the ability to use context to decipher words, as well as fluent and effortless reading. By fourth grade, children are expected to be efficient readers, reading rapidly, comprehending complex materials, and making inferences about the text [4,7,8]. A greater reliance on meaning is also evident. Characteristics of children in fourth grade and above include the ability to concentrate less on the print and more on details and ideas. This developmental shift has been referred to as a shift from learning to read to reading in order to learn .
The ultimate goal of reading development is efficient reading comprehension, defined as a process of extracting and constructing meaning through interaction and involvement with written language . What factors determine reading comprehension? Empirical evidence demonstrates that phonological processing, rapid automatized naming, orthographic processing, and word identification , as well as IQ , memory and attention , and higher order processes , all predict a significant portion of the variance in reading comprehension. However, even when these measures are entered into regression models, much of the variance in reading comprehension remains unexplained .
These findings have resulted in a shift towards a multi-dimensional view of reading comprehension that goes beyond cognitive and linguistic processes. The RAND model of reading comprehension  suggests that in order to understand the complex process of comprehension there is a need to concurrently examine a triangulation of the contribution of reader characteristics, text type, and environmental factors [13,15,8,9]. Interestingly, while much work has focused on the influence of the nature of the text type on reading (e.g., narrative vs. expository text) , very little work has focused on the typographical properties of text presentation. Could altering the perceptual features of the words, such as font size, line spacing, etc.’, actually lead to performance differences in reading comprehension? A central assumption in the reading comprehension literature is that in order to improve comprehension, the reader must improve his skills (e.g., phonological skill, vocabulary, decoding abilities, reading rate). However, what if comprehension can be improved by simply changing factors that are external to the reader, such as the typographical properties of the text, without changing its content? The current study was designed to address these questions.
The Effects of Manipulating Perceptual Presentation on Cognitive Performance
Many education researchers believe that reducing extraneous cognitive load is always beneficial for the learning situation. If a student was able to learn new information easily, both the student and the teacher are likely to label the session as successful regardless of whether the student is able to retrieve the information later . However, research in cognitive psychology suggests just the opposite. In many cases, the more challenging a learning session is, the better subsequent long term memory for the material studied in that session will be . It may be that greater cognitive engagement leads to deeper processing, which then facilitates encoding and subsequent retrieval . Thus, it has been found that the most effective learning strategies involve introducing difficulties for the learner. One clear example of a "desirable difficulty" [17,19] is the interleaving of to-be-learned materials, rather than blocking them, in a way that creates, at least temporarily, contextual interference for the learner . Interleaving has been found to produce stronger learning than blocking, at least in the long run.
Yet another way to make learning more challenging is to manipulate disfluency, the subjective metacognitive experience of difficulty associated with cognitive tasks. Disfluency has been found to be strongly related to confidence in the ability to remember new information , with greater disfluency yielding lower confidence. In turn, when learners are less confident in how well they have learned the material, they are more likely to engage in more effortful and elaborative processing . Indeed, disfluency has been shown to impact cognitive processing independently of actual cognitive difficulty (for example, the amount of material to be studied) . For example, a recent study has demonstrated that creating disfluency by presenting words upside-down for study enhanced later recall for these words, compared to words that were presented right-side up . It has also been shown that disfluency leads people to process information more carefully  and yields better oral comprehension . Based on the above findings, we raise the following questions: Can manipulating perceptual features of text, which have been shown to create disfluency effects in adults, also lead to better reading comprehension in children? Will the effects of the manipulation depend on stages of reading development?
The Developmental Effects of Manipulating Perceptual Presentation of Text
Manipulations of perceptual features of text build upon the assumption that the visual system makes use of relative size as a perceptual cue that conveys important information regarding the proximity of a stimulus . Oppenhiemer and his colleagues manipulated perceptual presentation of text simply by adopting fonts that were more difficult to read [1,28] by choosing faded shades, small fonts, and unclear photocopying of text. A different way to manipulate text presentation and create text disfluency may be to manipulate the spacing between lines and line lengths, under the assumption that these changes pose greater challenge for readers .
Manipulating text presentation may affect reading rate and accuracy  as well as feeling of proficiency. Indeed, studies focusing on feelings of proficiency, mainly in adults, report that processing words presented in larger fonts was subjectively more fluent than processing words presented in smaller fonts [30,31]. Importantly, it has been found that simple interventions, such as presenting educational materials on PowerPoint slides and handouts in italics (which children are less accustomed to) as opposed to presenting the same text in a standard, non-italesized format, engaged both university and high school students in more elaborative processing and even subsequently resulted in improved educational outcomes, including higher grades . In contrast, a study of word lists showed that font size did not affect recall [performance] among university students, though it did affect their judgment of learning, with larger fonts associated with greater estimations of remembrance . In these studies the participants were all skilled readers that were passed the initial phases of reading development. However, the effect of such manipulations may be different for poor readers as well as for younger children.
Studies of adults that have taken into account variability in reading skills have found that manipulating text presentation has opposite effects on good and poor readers. Thus, increasing text difficulty by deleting letters led poor readers to show decreased recall, whereas good readers showed improved recall with letter deletion . Based on these findings, we suggest that manipulating perceptual presentation of text might have differential effects on reading comprehension for skilled versus novice readers.
The majority of studies manipulating perceptual presentation of text in children have manipulated font size and line length and examined the effects on reading rate and accuracy and have not looked at its effect on reading comprehension. Interestingly, studies that examined font size found different effects for children at different stages of reading development: A relatively small font size was found to decrease the reading rates of five- to seven-year-olds, but had no effect on children in third to fifth grade . Similarly, another study compared the reading rates of children with dyslexia in second through fourth grade and reading-level matched controls . Dyslexic children benefited from larger fonts while their reading-level matched peers, similar to the results of college students previously described , showed no font size effects on reading rate and accuracy.
Regarding line length, in a study on six-year-olds, no differences were found in reading rate and accuracy between short and long lines, controlling for the number of words in a line . However, another study found that large fonts were read as well as smaller fonts with large spacing between the words (which results in longer lines) . Since there were no conclusive results across the two studies, and as line length and font size were concurrently manipulated in this study, conclusions cannot be drawn regarding each factor in relation to reading rate and accuracy.
Furthermore, the only study that looked at the effects of manipulating text presentation on reading comprehension, in children found that fonts with decorations (i.e., disfluent fonts) were comprehended as well as fonts without them . However, based on the findings of the effects of font size and line length on rate and accuracy of sentence reading, we would except they may also influence reading comprehension. Such information may have far reaching applied implications. As reading comprehension required the orchestration of many subskills , and an interaction between reader and text, it is important to study the effects of text presentation beyond the reading speed and accuracy level. Thus while rate and accuracy are necessary for comprehension, they are not sufficient. Factors that influence them may influence comprehension in a different manner.
To summarize, creating less accessible perceptual presentation of text, or disfluent text (smaller fonts, less spacing) was found to have different effects on the reading speed and accuracy of skilled versus unskilled readers. In terms of size, larger font size enhanced reading speed and accuracy of younger and dyslexic readers and showed no effect on older children. In addition, it did not affect recall in older university students. However, bolding or italicizing text did improve long-term memory in older high school and university students. To the extent that text presentation affects reading rate and accuracy, we would expect it to influence reading comprehension as well. Thus, we hypothesized that for younger readers, manipulating text presentation by increasing disfluency compared to the standard text they are used to would impede comprehension, as they still receive important contextual cues from the print. For older children, who have already mastered the decoding and efficiency stages and thus rely less on actual visual cues, we hypothesized that increased disfluency (less familiar and accessibly text presentation) would function as a desirable difficulty, resulting in deeper processing and thereby increasing comprehension.
The Current Experiments
In the current experiments, we examined the effect of perceptual fluency on reading comprehension in second and fifth grade children (Experiments 1 and 2, respectively). Specifically, we asked whether font size, line length, and line spacing would affect performance on a reading comprehension task. In addition, we asked whether these factors would differentially affect children in earlier versus later stages of reading development.
Experiment 1: Second Grade
Experiment 1 was designed to examine how the perceptual disfluency of text, created by decreasing font size, increasing line length, and decreasing line spacing, affects reading comprehension among second graders.
Participants were 45 second graders (20 girls, mean age 7.5 years) from elementary schools in Israel, mostly of middle- and upper-middle-class socioeconomic background. All children had rapid naming, reading and verbal abilities in the average range, based on standardized measures [39,40].
The research conducted in this paper was approved by the university review board - The Ethics Committee Review Board-IRB. The members of the Helsinki committee in our university are Shoshi Zalka and Avi Karni. Informed written consent was obtained from the parents and children, also, the data were analyzed anonymously). Finally, the investigation was conducted according to the principles expressed in the Declaration of Helsinki.
To examine reading comprehension, we developed a tool that included four age-appropriate texts, matched for level of difficulty and length. The texts were adapted from previous national reading assessment materials and were 44-47 words long. We manipulated three dimensions between the texts: font size, line length, and spacing between lines. The dimensions of the baseline text—20 pt font size, 4.2 inch line length, double line spacing—represented the text dimensions that are used for national reading assessment for the second grade, and reflected the typical font size, line length, and line spacing used in textbooks for this grade. For the other three texts, we manipulated presentation by decreasing font size in one text by 20%; increasing line length in another text by 20%; and decreasing line spacing by 20% in the final text. The assignment of each specific text to one of the four dimensions, as well as the order of text presentation, was counterbalanced across participants. For ecological reasons, texts were presented to children in a booklet. The design was self paced, as to ensure children are reading in their natural pace that is most comfortable for them.
After reading each text, students were asked to answer four multiple choice reading comprehension questions that were developed especially for the current study. Prior to conducting the study, the texts and questions were given to 10 judges, all with master’s degrees in literacy, to assess that they are indeed age-appropriate. The reliability of the tool, as examined in a pilot study with 51 children, was high (Cronbach’s α = .756).
The consent of the parents, the children, and of the school was obtained before beginning the study. Children were first tested individually in a quiet room at school to determine reading and vocabulary levels. Next, there was a group administration of the reading comprehension tool. Children were told they would be asked to read several passages and answer some questions about them. They had up to 30 minutes to complete the entire task.
For each participant, we calculated an overall reading comprehension score for each text by computing the proportion of reading comprehension questions answered correctly out of four. Results are presented in Figure 1. The analysis yielded a significant effect of font size on reading comprehension: second graders had higher comprehension scores on the standard font-size text (.87) than the small font-size text (.79), t(43) = 2.32, p< .05, Cohen’s d = .35. The analysis also yielded a significant effect of line length: Second graders had higher comprehension scores on the standard font-size text than the large line-length text (.74), t(40) = 3.35, p< .01, Cohen’s d = .54. Comparing reading comprehension between the standard text and the small spacing text did not yield a significant effect, t(43) = 1.00, ns, Cohen’s d = .153.
Experiment 2: Fifth Grade
The results of Experiment 1 support the prediction that for young children learning to read, increasing the perceptual disfluency of text, by decreasing font size or increasing line length, impairs comprehension. Experiment 2 was designed to examine the hypothesis that the opposite pattern, i.e., increased comprehension for more disfluent texts, would emerge for older children in fifth grade.
Participants were 45 fifth graders (24 girls, mean age 10.5 years) drawn from the same schools as the participants in Experiment 1. All children had reading and verbal abilities in the average range, based on standardized measures.
Materials and Procedure.
To examine reading comprehension, we developed a tool that was equivalent to the one used in Experiment 1 but adapted for fifth graders by employing the following changes: (1) the four age-appropriate texts were adapted from previous national reading assessment materials for the fifth grade and were 110-120 words long; (2) the dimensions of the baseline text represented the text dimensions that are used for national reading assessment for the fifth grade, and reflected the typical font size, line length, and line spacing used in textbooks for this grade—13 pt font size, 4.6 inch line length, one and a half line spacing. Again, disfluency was created for the other three texts by decreasing the font size of one text by 20%, increasing the line length of another text by 20%, and decreasing line spacing by 20% for the final text. The reliability of the tool, as examined in a pilot study with 50fifth graders, was high (Cronbach’s α = .797). The procedure was identical to the one used in Experiment 1.
For each participant, we calculated an overall reading comprehension score for each text by computing the proportion of reading comprehension questions answered correctly out of four. Results are presented in Figure 2. Results yielded a significant effect of font size on reading comprehension: in contrast to the effect found for second graders, fifth graders had higher comprehension scores for the smaller font-size text (.91) than the standard font-size text (.81), t(44) = -2.72, p< .01, Cohen’s d = .43. No significant effects of line length or line spacing were obtained, t(44) = .92, ns, Cohen’s d = .14 and t(44) = -.33, ns, Cohen’s d = .02, respectively.
Original article: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0074061